The invention discloses a heavy truck battery swap station two-stage battery charging optimization method based on deep reinforcement learning. The method comprises the following steps: S1, collecting daily operation data of a power grid and a battery swap station; s2, fitting the number of arrival batteries in each time period and historical data of SOE of the batteries in arrival by using a polynomial, and predicting battery replacement data in a future optimization time period; s3, establishing a two-stage ordered charging optimization model; and S4, carrying out training solution on the model by utilizing reinforcement learning. According to the invention, a two-stage ordered charging strategy is provided, according to the charging characteristics, charging is divided into multiple stages, and the charging time period is dynamically adjusted to adapt to different battery replacement requirements, so that the charging cost is reduced, the battery replacement efficiency is improved, and meanwhile, the overall use efficiency of the battery is improved.
一种基于深度强化学习的重卡换电站两段式电池充电优化方法,包括以下步骤:S1、收集收集电网和换电站的日常运营数据;S2、利用多项式对各时段到站电池的数量及到站时电池SOE的历史数据进行拟合并预测未来一个优化时段的换电数据;S3、建立两段式有序充电优化模型;S4、利用强化学习对模型进行训练求解。本发明提出一种两段式有序充电策略,根据其充电特性,将充电多段化,动态调整充电时间段,以适应不同的换电需求,从而减少充电成本,提高换电效率,同时提升电池的整体使用效率。
Heavy truck battery swap station two-stage battery charging optimization method based on deep reinforcement learning
一种基于深度强化学习的重卡换电站两段式电池充电优化方法
2024-09-17
Patent
Electronic Resource
Chinese
Heavy truck battery swap station and battery swap control method
European Patent Office | 2023
|Heavy truck battery swap station and battery swap control method
European Patent Office | 2022
|